Copilotkit Pitch Deck

CopilotKit Pitch Deck Wizard - Production

Purpose

Maintain and enhance the production-ready CopilotKit pitch deck wizard in the main Medellin Spark application. Uses CopilotKit Cloud API + Supabase Edge Functions (NOT LangGraph three-tier architecture).

Architecture Overview

Current Production Setup:

Frontend (:8080) --> Supabase Edge Functions --> OpenAI API
  React 19 + Vite     Deno Runtime (cloud)      GPT-4o-mini

Key Components:

• Frontend: src/pages/PitchDeckWizard.tsx - CopilotKit v1.10.6 with chat interface
• Edge Functions: supabase/functions/ - pitch-deck-assistant, generate-pitch-deck, chat
• Database: Supabase PostgreSQL with RLS enabled
• Status: 🟢 PRODUCTION READY (98/100)

Production Status

Tracker: /home/sk/template-copilot-kit-py/mvp-plan/progress/00-PRODUCTION-READINESS-TRACKER.md Latest Validation: /home/sk/template-copilot-kit-py/mvp-plan/progress/01-FINAL-PRODUCTION-VALIDATION.md

Current Implementation ✅

CopilotKit Integration (src/pages/PitchDeckWizard.tsx:189-192):

<CopilotKit
  publicApiKey={import.meta.env.VITE_COPILOT_CLOUD_PUBLIC_API_KEY}
  showDevConsole={import.meta.env.DEV}
>
  {/* Chat interface and wizard content */}
</CopilotKit>

Edge Functions (All ACTIVE):

1. chat (v16) - OpenAI proxy for secure API calls
2. pitch-deck-assistant (v26) - Conversation handler with progress tracking
3. generate-pitch-deck (v28) - Full deck generation (10 slides)
4. agent-example (v8) - Agent SDK demonstration

Features Working ✅:

• AI-powered chat interface
• Progress tracking (0-100%)
• Data collection via conversation
• Generate deck button at 80%+ completion
• Presentation viewer/editor
• Real-time updates

Performance Metrics

Build Performance:

Build Time: 4.71s ✅ (Target: <5s)
Bundle Size: 554KB ✅ (50% reduction from 1,109KB)
Gzip: 151KB ✅ (73% compression)
Warnings: 0 ✅

Production Quality:

TypeScript Errors: 0 ✅
Linter Warnings: 0 ✅
Routes Working: 24/24 ✅
Security Score: 100% ✅
Overall Score: 98/100 ✅

Common Issues & Solutions

Issue 1: Chat Interface Not Responding

Cause: Edge Function connection issue Check:

# Verify Edge Functions deployed
supabase functions list

# Check function logs
supabase functions logs pitch-deck-assistant --tail

Fix: Ensure all Edge Functions deployed and ACTIVE

Issue 2: API Key Errors

Cause: Missing or invalid OpenAI API key in Edge Function Check: Edge Function secrets

# View secrets (won't show values)
supabase secrets list

# Set OpenAI key if missing
supabase secrets set OPENAI_API_KEY=sk-...

Fix: Ensure OPENAI_API_KEY set as Supabase secret

Issue 3: Progress Bar Not Updating

Cause: Edge Function not returning completeness data Check: Browser console for response data Debug:

// In PitchDeckWizard.tsx - check response structure
console.log('Response from Edge Function:', response);
// Should include: { completeness: number, ... }

Issue 4: Build Errors

Cause: TypeScript errors or missing dependencies Check:

# Type check
pnpm tsc --noEmit

# Check for missing dependencies
pnpm install

Fix: Resolve TypeScript errors before building

Issue 5: Local Development Server Issues

Cause: Port conflicts or build cache Fix:

# Kill any processes on port 8080
lsof -ti :8080 | xargs kill -9

# Clear Vite cache
rm -rf node_modules/.vite

# Restart dev server
pnpm dev

Testing Workflow

1. Start Development Server

cd /home/sk/template-copilot-kit-py
pnpm dev

Expected:

VITE v7.x ready in 234 ms
➜  Local:   http://localhost:8080/

2. Navigate to Pitch Deck Wizard

URL: http://localhost:8080/pitch-deck-wizard

Check:

• ✅ Page loads without errors
• ✅ CopilotKit chat interface visible
• ✅ No console errors
• ✅ Dev console badge visible (in DEV mode)

3. Test Chat Flow

1. Send: "I want to create a pitch deck for TestCorp"
2. Verify: AI responds with questions
3. Send: "We're in AI software industry"
4. Verify: Conversation continues
5. Continue: Answer 4-5 more questions
6. Verify: Progress tracking works (if implemented)
7. Verify: "Generate Deck" appears at completion

4. Test Deck Generation

1. Click: "Generate Deck" button
2. Verify: Loading state shows
3. Wait: 5-10 seconds for generation
4. Verify: Redirect to /presentations/{id}/outline
5. Verify: All 10 slides render correctly

5. Production Build Test

# Build for production
pnpm build

# Should complete in <5s with 0 warnings
# Check output for bundle sizes

Key Files Reference

Frontend Files (Main Application)

• src/pages/PitchDeckWizard.tsx - CopilotKit integration and chat interface
• src/lib/apiClient.ts - Edge Function API calls
• package.json - Dependencies and build scripts
• vite.config.ts - Build configuration (optimized with code splitting)

Edge Function Files

• supabase/functions/chat/index.ts - OpenAI proxy (v16)
• supabase/functions/pitch-deck-assistant/index.ts - Conversation handler (v26)
• supabase/functions/generate-pitch-deck/index.ts - Deck generator (v28)
• supabase/functions/agent-example/index.ts - Agent SDK demo (v8)

Database Files

• supabase/migrations/ - Database schema migrations
• SQL tables: presentations, pitch_conversations, profiles

Documentation

• CLAUDE.md - Project standards and best practices
• mvp-plan/progress/00-PRODUCTION-READINESS-TRACKER.md - Production status
• mvp-plan/progress/01-FINAL-PRODUCTION-VALIDATION.md - Latest validation
• SITEMAP-2.md - Complete route map

Production Verification Checklist

Code Quality ✅

• TypeScript: 0 errors (pnpm tsc --noEmit)
• Linter: Clean (pnpm lint)
• No console.log in production code
• Error boundaries implemented
• Loading states added

CopilotKit Integration ✅

• CopilotKit v1.10.6 installed
• Public API key configured in .env
• Chat interface renders correctly
• Dev console enabled in DEV mode
• No CORS errors

Edge Functions ✅

• All 4 functions deployed and ACTIVE
• OPENAI_API_KEY set as secret
• Functions return proper responses
• Error handling comprehensive
• Logs accessible via supabase functions logs

Features ✅

• Chat conversation works end-to-end
• AI responds appropriately
• Data collection functional
• Generate deck button appears
• Deck generation creates 10 slides
• Presentation viewer/editor working

Build & Performance ✅

• Production build succeeds (<5s)
• Bundle optimized (554KB, 50% reduction)
• No build warnings
• Gzip compression enabled (151KB)
• Code splitting implemented

Quick Commands

Start Development:

cd /home/sk/template-copilot-kit-py
pnpm dev  # Runs on http://localhost:8080

Type Check:

pnpm tsc --noEmit  # Should return 0 errors

Production Build:

pnpm build  # Should complete in <5s with 0 warnings

Deploy Edge Functions:

# Deploy single function
supabase functions deploy chat

# List all functions
supabase functions list

# View logs
supabase functions logs pitch-deck-assistant --tail

Database Commands:

# Push migrations
npx supabase db push

# Check status
supabase status

Dependencies

Frontend (package.json):

• @copilotkit/react-core ^1.10.6 - CopilotKit SDK
• @copilotkit/react-ui ^1.10.6 - Chat interface
• react ^19.x
• vite ^7.x
• @supabase/supabase-js - Supabase client

Edge Functions (Deno runtime):

• OpenAI SDK for API calls
• Supabase client for database access
• Deno standard library

Environment Variables

Required in .env:

# CopilotKit (frontend)
VITE_COPILOT_CLOUD_PUBLIC_API_KEY=your_public_key

# Supabase (frontend)
VITE_SUPABASE_URL=https://your-project.supabase.co
VITE_SUPABASE_ANON_KEY=your_anon_key

# Edge Functions (Supabase secrets)
OPENAI_API_KEY=sk-...  # Set via: supabase secrets set

Performance Best Practices

✅ Implemented:

• Bundle code splitting (5 optimized chunks)
• Gzip compression (73% reduction)
• React.memo for expensive components
• Lazy loading for routes
• Error boundaries for graceful failures

🟡 Optional Enhancements:

• Add request debouncing for API calls
• Implement response caching
• Monitor Core Web Vitals
• Add service worker for offline support

Related Resources

• CopilotKit Docs: https://docs.copilotkit.ai
• Supabase Edge Functions: https://supabase.com/docs/guides/functions
• Project Documentation: /home/sk/template-copilot-kit-py/CLAUDE.md
• Production Tracker: /home/sk/template-copilot-kit-py/mvp-plan/progress/

Next Steps (Post-Production)

1. 🟡 Monitor production errors and Edge Function logs
2. 🟡 Collect user feedback on AI conversation quality
3. 🟡 Add analytics tracking for usage metrics
4. 🟡 Implement E2E test suite with Playwright
5. 🟡 Optimize prompts based on user interactions

This skill provides complete guidance for the production-ready CopilotKit pitch deck wizard in the main Medellin Spark application.